Step 3.5 Flash
stepfun/step-3.5-flash
fast sparse-MoE agentic reasoning.
cmd --model stepfun/step-3.5-flash
Intelligence index
26.5
Output speed
not yet scored
Input
$0.10 /M
Output
$0.30 /M
Cache read
$0.02 /M
Agent-loop cost
$0.04 /M in
Context window
1M tokens
Released
April 2, 2026
Modalities
→
vs. the lineup
Step 3.5 Flash beside its stablemates and nearest rivals. The ◆ marks the best value in each column across every row shown.
pin a rival:
| Model | Intelligence | Coding | Speed | Input $/M | Output $/M | Blended $/M | Context |
|---|---|---|---|---|---|---|---|
| Muse Spark 1.2 Contributor | 56.8◆ | 72.2◆ | —◆ | $0.10◆ | $0.20◆ | $0.13◆ | 1.05M◆ |
| MiniMax M2.5 | 34.5◆ | —◆ | —◆ | $0.30◆ | $1.20◆ | $0.52◆ | 200K◆ |
| Step 3.7 Flash | 30.9◆ | 39.6◆ | 391.3◆ | $0.20◆ | $1.15◆ | $0.44◆ | 256K◆ |
| Step 3.5 Flash ◆ | 26.5◆ | —◆ | —◆ | $0.10◆ | $0.30◆ | $0.15◆ | 1M◆ |
| Gemini 3.1 Flash Lite | 25.6◆ | 34.7◆ | —◆ | $0.25◆ | $1.50◆ | $0.56◆ | 1M◆ |
| Claude Haiku 4.5 | 24.1◆ | —◆ | 96.3◆ | $1◆ | $5◆ | $2◆ | 200K◆ |
| DeepSeek V4 Pro (latest)pinned | 45.3◆ | 59.4◆ | 63.3◆ | $0.66◆ | $1.98◆ | $0.99◆ | 1M◆ |
| DeepSeek V4 Flash (latest)pinned | 52◆ | 69.1◆ | 115.9◆ | $0.22◆ | $0.66◆ | $0.33◆ | 1M◆ |
coding performance
The Intelligence Index and its sub-scores, ranked against every scored model in the catalog. A metric that has not been measured for Step 3.5 Flash has been left empty.
IFBench
66.5
instruction following
Intelligence Index
26.5
#44 of 46 scored
Long-context reasoning
60.3
reasoning across a long context
SciCode
38.5
scientific coding
GPQA Diamond
82.6
graduate-level QA
usage calculator
How far a month of credits goes on Step 3.5 Flash.
Input tokensfresh prompt
800Output tokensmodel reply
180Cache read tokensre-read context
50Kcost / request $0.0011 · in $0.10 · out $0.30 · cache $0.02 per M
fresh input 7%output 5%cache reads 88%
Requests / 30 days
8.8K
$10 credits ÷ $0.0011 per request
~1.8K quick fixes~353 bug fixes~59 feature PRs
what real work costs
Real coding tasks priced end to end on Step 3.5 Flash, from a quick lookup to a full-repo agent run.
One agent task
$0.01
180K in at 75% cache hit, 12K out
What you pay
$0.06 /M
all-in across every token that task touched
Sticker input
$0.10 /M
cache reads bill at $0.02 /M instead
| Task | Tokens in · out | Step 3.5 Flash | Muse Spark 1.2 Contributor | Claude Haiku 4.5 |
|---|---|---|---|---|
| Quick lookup / one-liner | 8K · 1K | $0.0005 | $0.0003 | $0.0065 |
| Review a 500-line PR | 60K · 4K | $0.0038 | $0.0027 | $0.04 |
| Fix a bug (agent loop) | 180K · 12K | $0.01 | $0.0072 | $0.12 |
| Refactor a module | 320K · 20K | $0.02 | $0.01 | $0.21 |
| Full-repo agent run | 900K · 45K | $0.05 | $0.03 | $0.49 |
frequently asked
What is Step 3.5 Flash best for?
Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token. It is a reasoning model that is incredibly speed efficient even at long contexts.
How much does Step 3.5 Flash cost?
$0.10/M input and $0.30/M output, cache reads $0.02/M. In an agent loop most input is cache-read, so the effective input rate is about $0.04/M.Which plan do I need?
Available on Go and above.
How do I switch to it?
Run
cmd --model stepfun/step-3.5-flash, or type /model in a session and pick it. You can switch mid-session without losing context.Ship code that matches your taste
Command Code is the AI coding agent that continuously learns your taste. Start for $1.
Benchmarks from Artificial Analysis (v4.1)commandcode.ai/models/step-3-5-flash